From 8287889742940cf3c416e755322090d09f2829be Mon Sep 17 00:00:00 2001 From: Dave Chinner Date: Sat, 26 Mar 2011 09:13:08 +1100 Subject: [PATCH 1/6] xfs: preallocation transactions do not need to be synchronous Preallocation and hole punch transactions are currently synchronous and this is causing performance problems in some cases. The transactions don't need to be synchronous as we don't need to guarantee the preallocation is persistent on disk until a fdatasync, fsync, sync operation occurs. If the file is opened O_SYNC or O_DATASYNC, only then should the transaction be issued synchronously. Signed-off-by: Dave Chinner Reviewed-by: Christoph Hellwig Reviewed-by: Alex Elder --- fs/xfs/linux-2.6/xfs_file.c | 6 +++++- fs/xfs/linux-2.6/xfs_ioctl.c | 4 ++++ fs/xfs/xfs_vnodeops.c | 3 ++- fs/xfs/xfs_vnodeops.h | 1 + 4 files changed, 12 insertions(+), 2 deletions(-) diff --git a/fs/xfs/linux-2.6/xfs_file.c b/fs/xfs/linux-2.6/xfs_file.c index a55c1b46b219..52aadfbed132 100644 --- a/fs/xfs/linux-2.6/xfs_file.c +++ b/fs/xfs/linux-2.6/xfs_file.c @@ -896,6 +896,7 @@ xfs_file_fallocate( xfs_flock64_t bf; xfs_inode_t *ip = XFS_I(inode); int cmd = XFS_IOC_RESVSP; + int attr_flags = XFS_ATTR_NOLOCK; if (mode & ~(FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE)) return -EOPNOTSUPP; @@ -918,7 +919,10 @@ xfs_file_fallocate( goto out_unlock; } - error = -xfs_change_file_space(ip, cmd, &bf, 0, XFS_ATTR_NOLOCK); + if (file->f_flags & O_DSYNC) + attr_flags |= XFS_ATTR_SYNC; + + error = -xfs_change_file_space(ip, cmd, &bf, 0, attr_flags); if (error) goto out_unlock; diff --git a/fs/xfs/linux-2.6/xfs_ioctl.c b/fs/xfs/linux-2.6/xfs_ioctl.c index 0ca0e3c024d7..acca2c5ca3fa 100644 --- a/fs/xfs/linux-2.6/xfs_ioctl.c +++ b/fs/xfs/linux-2.6/xfs_ioctl.c @@ -624,6 +624,10 @@ xfs_ioc_space( if (filp->f_flags & (O_NDELAY|O_NONBLOCK)) attr_flags |= XFS_ATTR_NONBLOCK; + + if (filp->f_flags & O_DSYNC) + attr_flags |= XFS_ATTR_SYNC; + if (ioflags & IO_INVIS) attr_flags |= XFS_ATTR_DMI; diff --git a/fs/xfs/xfs_vnodeops.c b/fs/xfs/xfs_vnodeops.c index 37d8146ee15b..c48b4217ec47 100644 --- a/fs/xfs/xfs_vnodeops.c +++ b/fs/xfs/xfs_vnodeops.c @@ -2831,7 +2831,8 @@ xfs_change_file_space( ip->i_d.di_flags &= ~XFS_DIFLAG_PREALLOC; xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE); - xfs_trans_set_sync(tp); + if (attr_flags & XFS_ATTR_SYNC) + xfs_trans_set_sync(tp); error = xfs_trans_commit(tp, 0); diff --git a/fs/xfs/xfs_vnodeops.h b/fs/xfs/xfs_vnodeops.h index f6702927eee4..3bcd23353d6c 100644 --- a/fs/xfs/xfs_vnodeops.h +++ b/fs/xfs/xfs_vnodeops.h @@ -18,6 +18,7 @@ int xfs_setattr(struct xfs_inode *ip, struct iattr *vap, int flags); #define XFS_ATTR_NONBLOCK 0x02 /* return EAGAIN if operation would block */ #define XFS_ATTR_NOLOCK 0x04 /* Don't grab any conflicting locks */ #define XFS_ATTR_NOACL 0x08 /* Don't call xfs_acl_chmod */ +#define XFS_ATTR_SYNC 0x10 /* synchronous operation required */ int xfs_readlink(struct xfs_inode *ip, char *link); int xfs_release(struct xfs_inode *ip); From a19fb380961f209a3a406443686647bcd01bb9a6 Mon Sep 17 00:00:00 2001 From: Dave Chinner Date: Sat, 26 Mar 2011 09:13:42 +1100 Subject: [PATCH 2/6] vmap: flush vmap aliases when mapping fails On 32 bit systems, vmalloc space is limited and XFS can chew through it quickly as the vmalloc space is lazily freed. This can result in failure to map buffers, even when there is apparently large amounts of vmalloc space available. Hence, if we fail to map a buffer, purge the aliases that have not yet been freed to hopefuly free up enough vmalloc space to allow a retry to succeed. Signed-off-by: Dave Chinner Reviewed-by: Christoph Hellwig Reviewed-by: Alex Elder --- fs/xfs/linux-2.6/xfs_buf.c | 14 +++++++++++--- 1 file changed, 11 insertions(+), 3 deletions(-) diff --git a/fs/xfs/linux-2.6/xfs_buf.c b/fs/xfs/linux-2.6/xfs_buf.c index 3cc671c8a67f..a5a260fab824 100644 --- a/fs/xfs/linux-2.6/xfs_buf.c +++ b/fs/xfs/linux-2.6/xfs_buf.c @@ -455,9 +455,17 @@ _xfs_buf_map_pages( bp->b_addr = page_address(bp->b_pages[0]) + bp->b_offset; bp->b_flags |= XBF_MAPPED; } else if (flags & XBF_MAPPED) { - bp->b_addr = vm_map_ram(bp->b_pages, bp->b_page_count, - -1, PAGE_KERNEL); - if (unlikely(bp->b_addr == NULL)) + int retried = 0; + + do { + bp->b_addr = vm_map_ram(bp->b_pages, bp->b_page_count, + -1, PAGE_KERNEL); + if (bp->b_addr) + break; + vm_unmap_aliases(); + } while (retried++ <= 1); + + if (!bp->b_addr) return -ENOMEM; bp->b_addr += bp->b_offset; bp->b_flags |= XBF_MAPPED; From 1bfd8d04190c615bb8d1d98188dead0c09702208 Mon Sep 17 00:00:00 2001 From: Dave Chinner Date: Sat, 26 Mar 2011 09:13:55 +1100 Subject: [PATCH 3/6] xfs: introduce inode cluster buffer trylocks for xfs_iflush There is an ABBA deadlock between synchronous inode flushing in xfs_reclaim_inode and xfs_icluster_free. xfs_icluster_free locks the buffer, then takes inode ilocks, whilst synchronous reclaim takes the ilock followed by the buffer lock in xfs_iflush(). To avoid this deadlock, separate the inode cluster buffer locking semantics from the synchronous inode flush semantics, allowing callers to attempt to lock the buffer but still issue synchronous IO if it can get the buffer. This requires xfs_iflush() calls that currently use non-blocking semantics to pass SYNC_TRYLOCK rather than 0 as the flags parameter. This allows xfs_reclaim_inode to avoid the deadlock on the buffer lock and detect the failure so that it can drop the inode ilock and restart the reclaim attempt on the inode. This allows xfs_ifree_cluster to obtain the inode lock, mark the inode stale and release it and hence defuse the deadlock situation. It also has the pleasant side effect of avoiding IO in xfs_reclaim_inode when it tries to next reclaim the inode as it is now marked stale. Signed-off-by: Dave Chinner Reviewed-by: Christoph Hellwig Reviewed-by: Alex Elder --- fs/xfs/linux-2.6/xfs_super.c | 2 +- fs/xfs/linux-2.6/xfs_sync.c | 30 +++++++++++++++++++++++++++--- fs/xfs/xfs_inode.c | 2 +- fs/xfs/xfs_inode_item.c | 6 +++--- 4 files changed, 32 insertions(+), 8 deletions(-) diff --git a/fs/xfs/linux-2.6/xfs_super.c b/fs/xfs/linux-2.6/xfs_super.c index 818c4cf2de86..8a70b2a17d6f 100644 --- a/fs/xfs/linux-2.6/xfs_super.c +++ b/fs/xfs/linux-2.6/xfs_super.c @@ -1078,7 +1078,7 @@ xfs_fs_write_inode( error = 0; goto out_unlock; } - error = xfs_iflush(ip, 0); + error = xfs_iflush(ip, SYNC_TRYLOCK); } out_unlock: diff --git a/fs/xfs/linux-2.6/xfs_sync.c b/fs/xfs/linux-2.6/xfs_sync.c index 6c10f1d2e3d3..594cd822d84d 100644 --- a/fs/xfs/linux-2.6/xfs_sync.c +++ b/fs/xfs/linux-2.6/xfs_sync.c @@ -761,8 +761,10 @@ xfs_reclaim_inode( struct xfs_perag *pag, int sync_mode) { - int error = 0; + int error; +restart: + error = 0; xfs_ilock(ip, XFS_ILOCK_EXCL); if (!xfs_iflock_nowait(ip)) { if (!(sync_mode & SYNC_WAIT)) @@ -788,9 +790,31 @@ xfs_reclaim_inode( if (xfs_inode_clean(ip)) goto reclaim; - /* Now we have an inode that needs flushing */ - error = xfs_iflush(ip, sync_mode); + /* + * Now we have an inode that needs flushing. + * + * We do a nonblocking flush here even if we are doing a SYNC_WAIT + * reclaim as we can deadlock with inode cluster removal. + * xfs_ifree_cluster() can lock the inode buffer before it locks the + * ip->i_lock, and we are doing the exact opposite here. As a result, + * doing a blocking xfs_itobp() to get the cluster buffer will result + * in an ABBA deadlock with xfs_ifree_cluster(). + * + * As xfs_ifree_cluser() must gather all inodes that are active in the + * cache to mark them stale, if we hit this case we don't actually want + * to do IO here - we want the inode marked stale so we can simply + * reclaim it. Hence if we get an EAGAIN error on a SYNC_WAIT flush, + * just unlock the inode, back off and try again. Hopefully the next + * pass through will see the stale flag set on the inode. + */ + error = xfs_iflush(ip, SYNC_TRYLOCK | sync_mode); if (sync_mode & SYNC_WAIT) { + if (error == EAGAIN) { + xfs_iunlock(ip, XFS_ILOCK_EXCL); + /* backoff longer than in xfs_ifree_cluster */ + delay(2); + goto restart; + } xfs_iflock(ip); goto reclaim; } diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c index da871f532236..742c8330994a 100644 --- a/fs/xfs/xfs_inode.c +++ b/fs/xfs/xfs_inode.c @@ -2835,7 +2835,7 @@ xfs_iflush( * Get the buffer containing the on-disk inode. */ error = xfs_itobp(mp, NULL, ip, &dip, &bp, - (flags & SYNC_WAIT) ? XBF_LOCK : XBF_TRYLOCK); + (flags & SYNC_TRYLOCK) ? XBF_TRYLOCK : XBF_LOCK); if (error || !bp) { xfs_ifunlock(ip); return error; diff --git a/fs/xfs/xfs_inode_item.c b/fs/xfs/xfs_inode_item.c index fd4f398bd6f1..46cc40131d4a 100644 --- a/fs/xfs/xfs_inode_item.c +++ b/fs/xfs/xfs_inode_item.c @@ -760,11 +760,11 @@ xfs_inode_item_push( * Push the inode to it's backing buffer. This will not remove the * inode from the AIL - a further push will be required to trigger a * buffer push. However, this allows all the dirty inodes to be pushed - * to the buffer before it is pushed to disk. THe buffer IO completion - * will pull th einode from the AIL, mark it clean and unlock the flush + * to the buffer before it is pushed to disk. The buffer IO completion + * will pull the inode from the AIL, mark it clean and unlock the flush * lock. */ - (void) xfs_iflush(ip, 0); + (void) xfs_iflush(ip, SYNC_TRYLOCK); xfs_iunlock(ip, XFS_ILOCK_SHARED); } From 7401aafd5019d32a888e5f27332cf580945574bf Mon Sep 17 00:00:00 2001 From: Dave Chinner Date: Sat, 26 Mar 2011 09:14:44 +1100 Subject: [PATCH 4/6] xfs: xfs_trans_read_buf() should return an error on failure When inside a transaction and we fail to read a buffer, xfs_trans_read_buf returns a null buffer pointer and no error. xfs_do_da_buf() checks the error return, but not the buffer, and as a result this read failure condition causes a panic when it attempts to dereference the non-existant buffer. Make xfs_trans_read_buf() return the same error for this situation regardless of whether it is in a transaction or not. This means every caller does not need to check both the error return and the buffer before proceeding to use the buffer. Signed-off-by: Dave Chinner Reviewed-by: Alex Elder --- fs/xfs/xfs_trans_buf.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/fs/xfs/xfs_trans_buf.c b/fs/xfs/xfs_trans_buf.c index 3bea66132334..03b3b7f85a3b 100644 --- a/fs/xfs/xfs_trans_buf.c +++ b/fs/xfs/xfs_trans_buf.c @@ -383,7 +383,8 @@ xfs_trans_read_buf( bp = xfs_buf_read(target, blkno, len, flags | XBF_DONT_BLOCK); if (bp == NULL) { *bpp = NULL; - return 0; + return (flags & XBF_TRYLOCK) ? + 0 : XFS_ERROR(ENOMEM); } if (XFS_BUF_GETERROR(bp) != 0) { XFS_BUF_SUPER_STALE(bp); From 704b2907c2d47ceb187c0e25a6bbc2174b198f2f Mon Sep 17 00:00:00 2001 From: Dave Chinner Date: Sat, 26 Mar 2011 09:14:57 +1100 Subject: [PATCH 5/6] xfs: register the inode cache shrinker before quotachecks During mount, we can do a quotacheck that involves a bulkstat pass on all inodes. If there are more inodes in the filesystem than can be held in memory, we require the inode cache shrinker to run to ensure that we don't run out of memory. Unfortunately, the inode cache shrinker is not registered until we get to the end of the superblock setup process, which is after a quotacheck is run if it is needed. Hence we need to register the inode cache shrinker earlier in the mount process so that we don't OOM during mount. This requires that we also initialise the syncd work before we register the shrinker, so we nee dto juggle that around as well. While there, make sure that we have set up the block sizes in the VFS superblock correctly before the quotacheck is run so that any inodes that are cached as a result of the quotacheck have their block size fields set up correctly. Cc: stable@kernel.org Signed-off-by: Dave Chinner Reviewed-by: Alex Elder --- fs/xfs/linux-2.6/xfs_super.c | 34 ++++++++++++++++++++++++---------- 1 file changed, 24 insertions(+), 10 deletions(-) diff --git a/fs/xfs/linux-2.6/xfs_super.c b/fs/xfs/linux-2.6/xfs_super.c index 8a70b2a17d6f..1ba5c451da36 100644 --- a/fs/xfs/linux-2.6/xfs_super.c +++ b/fs/xfs/linux-2.6/xfs_super.c @@ -1539,10 +1539,14 @@ xfs_fs_fill_super( if (error) goto out_free_sb; - error = xfs_mountfs(mp); - if (error) - goto out_filestream_unmount; - + /* + * we must configure the block size in the superblock before we run the + * full mount process as the mount process can lookup and cache inodes. + * For the same reason we must also initialise the syncd and register + * the inode cache shrinker so that inodes can be reclaimed during + * operations like a quotacheck that iterate all inodes in the + * filesystem. + */ sb->s_magic = XFS_SB_MAGIC; sb->s_blocksize = mp->m_sb.sb_blocksize; sb->s_blocksize_bits = ffs(sb->s_blocksize) - 1; @@ -1550,6 +1554,16 @@ xfs_fs_fill_super( sb->s_time_gran = 1; set_posix_acl_flag(sb); + error = xfs_syncd_init(mp); + if (error) + goto out_filestream_unmount; + + xfs_inode_shrinker_register(mp); + + error = xfs_mountfs(mp); + if (error) + goto out_syncd_stop; + root = igrab(VFS_I(mp->m_rootip)); if (!root) { error = ENOENT; @@ -1565,14 +1579,11 @@ xfs_fs_fill_super( goto fail_vnrele; } - error = xfs_syncd_init(mp); - if (error) - goto fail_vnrele; - - xfs_inode_shrinker_register(mp); - return 0; + out_syncd_stop: + xfs_inode_shrinker_unregister(mp); + xfs_syncd_stop(mp); out_filestream_unmount: xfs_filestream_unmount(mp); out_free_sb: @@ -1596,6 +1607,9 @@ xfs_fs_fill_super( } fail_unmount: + xfs_inode_shrinker_unregister(mp); + xfs_syncd_stop(mp); + /* * Blow away any referenced inode in the filestreams cache. * This can and will cause log traffic as inodes go inactive From 0e6e847ffe37436e331c132639f9f872febce82e Mon Sep 17 00:00:00 2001 From: Dave Chinner Date: Sat, 26 Mar 2011 09:16:45 +1100 Subject: [PATCH 6/6] xfs: stop using the page cache to back the buffer cache Now that the buffer cache has it's own LRU, we do not need to use the page cache to provide persistent caching and reclaim infrastructure. Convert the buffer cache to use alloc_pages() instead of the page cache. This will remove all the overhead of page cache management from setup and teardown of the buffers, as well as needing to mark pages accessed as we find buffers in the buffer cache. By avoiding the page cache, we also remove the need to keep state in the page_private(page) field for persistant storage across buffer free/buffer rebuild and so all that code can be removed. This also fixes the long-standing problem of not having enough bits in the page_private field to track all the state needed for a 512 sector/64k page setup. It also removes the need for page locking during reads as the pages are unique to the buffer and nobody else will be attempting to access them. Finally, it removes the buftarg address space lock as a point of global contention on workloads that allocate and free buffers quickly such as when creating or removing large numbers of inodes in parallel. This remove the 16TB limit on filesystem size on 32 bit machines as the page index (32 bit) is no longer used for lookups of metadata buffers - the buffer cache is now solely indexed by disk address which is stored in a 64 bit field in the buffer. Signed-off-by: Dave Chinner Reviewed-by: Alex Elder --- fs/xfs/linux-2.6/xfs_buf.c | 341 +++++++++---------------------------- fs/xfs/linux-2.6/xfs_buf.h | 40 +---- 2 files changed, 84 insertions(+), 297 deletions(-) diff --git a/fs/xfs/linux-2.6/xfs_buf.c b/fs/xfs/linux-2.6/xfs_buf.c index a5a260fab824..d45b2cdee6c4 100644 --- a/fs/xfs/linux-2.6/xfs_buf.c +++ b/fs/xfs/linux-2.6/xfs_buf.c @@ -93,75 +93,6 @@ xfs_buf_vmap_len( return (bp->b_page_count * PAGE_SIZE) - bp->b_offset; } -/* - * Page Region interfaces. - * - * For pages in filesystems where the blocksize is smaller than the - * pagesize, we use the page->private field (long) to hold a bitmap - * of uptodate regions within the page. - * - * Each such region is "bytes per page / bits per long" bytes long. - * - * NBPPR == number-of-bytes-per-page-region - * BTOPR == bytes-to-page-region (rounded up) - * BTOPRT == bytes-to-page-region-truncated (rounded down) - */ -#if (BITS_PER_LONG == 32) -#define PRSHIFT (PAGE_CACHE_SHIFT - 5) /* (32 == 1<<5) */ -#elif (BITS_PER_LONG == 64) -#define PRSHIFT (PAGE_CACHE_SHIFT - 6) /* (64 == 1<<6) */ -#else -#error BITS_PER_LONG must be 32 or 64 -#endif -#define NBPPR (PAGE_CACHE_SIZE/BITS_PER_LONG) -#define BTOPR(b) (((unsigned int)(b) + (NBPPR - 1)) >> PRSHIFT) -#define BTOPRT(b) (((unsigned int)(b) >> PRSHIFT)) - -STATIC unsigned long -page_region_mask( - size_t offset, - size_t length) -{ - unsigned long mask; - int first, final; - - first = BTOPR(offset); - final = BTOPRT(offset + length - 1); - first = min(first, final); - - mask = ~0UL; - mask <<= BITS_PER_LONG - (final - first); - mask >>= BITS_PER_LONG - (final); - - ASSERT(offset + length <= PAGE_CACHE_SIZE); - ASSERT((final - first) < BITS_PER_LONG && (final - first) >= 0); - - return mask; -} - -STATIC void -set_page_region( - struct page *page, - size_t offset, - size_t length) -{ - set_page_private(page, - page_private(page) | page_region_mask(offset, length)); - if (page_private(page) == ~0UL) - SetPageUptodate(page); -} - -STATIC int -test_page_region( - struct page *page, - size_t offset, - size_t length) -{ - unsigned long mask = page_region_mask(offset, length); - - return (mask && (page_private(page) & mask) == mask); -} - /* * xfs_buf_lru_add - add a buffer to the LRU. * @@ -332,7 +263,7 @@ xfs_buf_free( ASSERT(list_empty(&bp->b_lru)); - if (bp->b_flags & (_XBF_PAGE_CACHE|_XBF_PAGES)) { + if (bp->b_flags & _XBF_PAGES) { uint i; if (xfs_buf_is_vmapped(bp)) @@ -342,25 +273,22 @@ xfs_buf_free( for (i = 0; i < bp->b_page_count; i++) { struct page *page = bp->b_pages[i]; - if (bp->b_flags & _XBF_PAGE_CACHE) - ASSERT(!PagePrivate(page)); - page_cache_release(page); + __free_page(page); } - } + } else if (bp->b_flags & _XBF_KMEM) + kmem_free(bp->b_addr); _xfs_buf_free_pages(bp); xfs_buf_deallocate(bp); } /* - * Finds all pages for buffer in question and builds it's page list. + * Allocates all the pages for buffer in question and builds it's page list. */ STATIC int -_xfs_buf_lookup_pages( +xfs_buf_allocate_memory( xfs_buf_t *bp, uint flags) { - struct address_space *mapping = bp->b_target->bt_mapping; - size_t blocksize = bp->b_target->bt_bsize; size_t size = bp->b_count_desired; size_t nbytes, offset; gfp_t gfp_mask = xb_to_gfp(flags); @@ -369,29 +297,55 @@ _xfs_buf_lookup_pages( xfs_off_t end; int error; + /* + * for buffers that are contained within a single page, just allocate + * the memory from the heap - there's no need for the complexity of + * page arrays to keep allocation down to order 0. + */ + if (bp->b_buffer_length < PAGE_SIZE) { + bp->b_addr = kmem_alloc(bp->b_buffer_length, xb_to_km(flags)); + if (!bp->b_addr) { + /* low memory - use alloc_page loop instead */ + goto use_alloc_page; + } + + if (((unsigned long)(bp->b_addr + bp->b_buffer_length - 1) & + PAGE_MASK) != + ((unsigned long)bp->b_addr & PAGE_MASK)) { + /* b_addr spans two pages - use alloc_page instead */ + kmem_free(bp->b_addr); + bp->b_addr = NULL; + goto use_alloc_page; + } + bp->b_offset = offset_in_page(bp->b_addr); + bp->b_pages = bp->b_page_array; + bp->b_pages[0] = virt_to_page(bp->b_addr); + bp->b_page_count = 1; + bp->b_flags |= XBF_MAPPED | _XBF_KMEM; + return 0; + } + +use_alloc_page: end = bp->b_file_offset + bp->b_buffer_length; page_count = xfs_buf_btoc(end) - xfs_buf_btoct(bp->b_file_offset); - error = _xfs_buf_get_pages(bp, page_count, flags); if (unlikely(error)) return error; - bp->b_flags |= _XBF_PAGE_CACHE; offset = bp->b_offset; - first = bp->b_file_offset >> PAGE_CACHE_SHIFT; + first = bp->b_file_offset >> PAGE_SHIFT; + bp->b_flags |= _XBF_PAGES; for (i = 0; i < bp->b_page_count; i++) { struct page *page; uint retries = 0; - - retry: - page = find_or_create_page(mapping, first + i, gfp_mask); +retry: + page = alloc_page(gfp_mask); if (unlikely(page == NULL)) { if (flags & XBF_READ_AHEAD) { bp->b_page_count = i; - for (i = 0; i < bp->b_page_count; i++) - unlock_page(bp->b_pages[i]); - return -ENOMEM; + error = ENOMEM; + goto out_free_pages; } /* @@ -412,33 +366,16 @@ _xfs_buf_lookup_pages( XFS_STATS_INC(xb_page_found); - nbytes = min_t(size_t, size, PAGE_CACHE_SIZE - offset); + nbytes = min_t(size_t, size, PAGE_SIZE - offset); size -= nbytes; - - ASSERT(!PagePrivate(page)); - if (!PageUptodate(page)) { - page_count--; - if (blocksize >= PAGE_CACHE_SIZE) { - if (flags & XBF_READ) - bp->b_flags |= _XBF_PAGE_LOCKED; - } else if (!PagePrivate(page)) { - if (test_page_region(page, offset, nbytes)) - page_count++; - } - } - bp->b_pages[i] = page; offset = 0; } + return 0; - if (!(bp->b_flags & _XBF_PAGE_LOCKED)) { - for (i = 0; i < bp->b_page_count; i++) - unlock_page(bp->b_pages[i]); - } - - if (page_count == bp->b_page_count) - bp->b_flags |= XBF_DONE; - +out_free_pages: + for (i = 0; i < bp->b_page_count; i++) + __free_page(bp->b_pages[i]); return error; } @@ -450,8 +387,9 @@ _xfs_buf_map_pages( xfs_buf_t *bp, uint flags) { - /* A single page buffer is always mappable */ + ASSERT(bp->b_flags & _XBF_PAGES); if (bp->b_page_count == 1) { + /* A single page buffer is always mappable */ bp->b_addr = page_address(bp->b_pages[0]) + bp->b_offset; bp->b_flags |= XBF_MAPPED; } else if (flags & XBF_MAPPED) { @@ -576,9 +514,14 @@ found: } } + /* + * if the buffer is stale, clear all the external state associated with + * it. We need to keep flags such as how we allocated the buffer memory + * intact here. + */ if (bp->b_flags & XBF_STALE) { ASSERT((bp->b_flags & _XBF_DELWRI_Q) == 0); - bp->b_flags &= XBF_MAPPED; + bp->b_flags &= XBF_MAPPED | _XBF_KMEM | _XBF_PAGES; } trace_xfs_buf_find(bp, flags, _RET_IP_); @@ -599,7 +542,7 @@ xfs_buf_get( xfs_buf_flags_t flags) { xfs_buf_t *bp, *new_bp; - int error = 0, i; + int error = 0; new_bp = xfs_buf_allocate(flags); if (unlikely(!new_bp)) @@ -607,7 +550,7 @@ xfs_buf_get( bp = _xfs_buf_find(target, ioff, isize, flags, new_bp); if (bp == new_bp) { - error = _xfs_buf_lookup_pages(bp, flags); + error = xfs_buf_allocate_memory(bp, flags); if (error) goto no_buffer; } else { @@ -616,9 +559,6 @@ xfs_buf_get( return NULL; } - for (i = 0; i < bp->b_page_count; i++) - mark_page_accessed(bp->b_pages[i]); - if (!(bp->b_flags & XBF_MAPPED)) { error = _xfs_buf_map_pages(bp, flags); if (unlikely(error)) { @@ -719,8 +659,7 @@ xfs_buf_readahead( { struct backing_dev_info *bdi; - bdi = target->bt_mapping->backing_dev_info; - if (bdi_read_congested(bdi)) + if (bdi_read_congested(target->bt_bdi)) return; xfs_buf_read(target, ioff, isize, @@ -798,10 +737,10 @@ xfs_buf_associate_memory( size_t buflen; int page_count; - pageaddr = (unsigned long)mem & PAGE_CACHE_MASK; + pageaddr = (unsigned long)mem & PAGE_MASK; offset = (unsigned long)mem - pageaddr; - buflen = PAGE_CACHE_ALIGN(len + offset); - page_count = buflen >> PAGE_CACHE_SHIFT; + buflen = PAGE_ALIGN(len + offset); + page_count = buflen >> PAGE_SHIFT; /* Free any previous set of page pointers */ if (bp->b_pages) @@ -818,13 +757,12 @@ xfs_buf_associate_memory( for (i = 0; i < bp->b_page_count; i++) { bp->b_pages[i] = mem_to_page((void *)pageaddr); - pageaddr += PAGE_CACHE_SIZE; + pageaddr += PAGE_SIZE; } bp->b_count_desired = len; bp->b_buffer_length = buflen; bp->b_flags |= XBF_MAPPED; - bp->b_flags &= ~_XBF_PAGE_LOCKED; return 0; } @@ -931,20 +869,7 @@ xfs_buf_rele( /* - * Mutual exclusion on buffers. Locking model: - * - * Buffers associated with inodes for which buffer locking - * is not enabled are not protected by semaphores, and are - * assumed to be exclusively owned by the caller. There is a - * spinlock in the buffer, used by the caller when concurrent - * access is possible. - */ - -/* - * Locks a buffer object, if it is not already locked. Note that this in - * no way locks the underlying pages, so it is only useful for - * synchronizing concurrent use of buffer objects, not for synchronizing - * independent access to the underlying pages. + * Lock a buffer object, if it is not already locked. * * If we come across a stale, pinned, locked buffer, we know that we are * being asked to lock a buffer that has been reallocated. Because it is @@ -978,10 +903,7 @@ xfs_buf_lock_value( } /* - * Locks a buffer object. - * Note that this in no way locks the underlying pages, so it is only - * useful for synchronizing concurrent use of buffer objects, not for - * synchronizing independent access to the underlying pages. + * Lock a buffer object. * * If we come across a stale, pinned, locked buffer, we know that we * are being asked to lock a buffer that has been reallocated. Because @@ -998,7 +920,7 @@ xfs_buf_lock( if (atomic_read(&bp->b_pin_count) && (bp->b_flags & XBF_STALE)) xfs_log_force(bp->b_target->bt_mount, 0); if (atomic_read(&bp->b_io_remaining)) - blk_run_address_space(bp->b_target->bt_mapping); + blk_run_backing_dev(bp->b_target->bt_bdi, NULL); down(&bp->b_sema); XB_SET_OWNER(bp); @@ -1043,7 +965,7 @@ xfs_buf_wait_unpin( if (atomic_read(&bp->b_pin_count) == 0) break; if (atomic_read(&bp->b_io_remaining)) - blk_run_address_space(bp->b_target->bt_mapping); + blk_run_backing_dev(bp->b_target->bt_bdi, NULL); schedule(); } remove_wait_queue(&bp->b_waiters, &wait); @@ -1256,10 +1178,8 @@ _xfs_buf_ioend( xfs_buf_t *bp, int schedule) { - if (atomic_dec_and_test(&bp->b_io_remaining) == 1) { - bp->b_flags &= ~_XBF_PAGE_LOCKED; + if (atomic_dec_and_test(&bp->b_io_remaining) == 1) xfs_buf_ioend(bp, schedule); - } } STATIC void @@ -1268,35 +1188,12 @@ xfs_buf_bio_end_io( int error) { xfs_buf_t *bp = (xfs_buf_t *)bio->bi_private; - unsigned int blocksize = bp->b_target->bt_bsize; - struct bio_vec *bvec = bio->bi_io_vec + bio->bi_vcnt - 1; xfs_buf_ioerror(bp, -error); if (!error && xfs_buf_is_vmapped(bp) && (bp->b_flags & XBF_READ)) invalidate_kernel_vmap_range(bp->b_addr, xfs_buf_vmap_len(bp)); - do { - struct page *page = bvec->bv_page; - - ASSERT(!PagePrivate(page)); - if (unlikely(bp->b_error)) { - if (bp->b_flags & XBF_READ) - ClearPageUptodate(page); - } else if (blocksize >= PAGE_CACHE_SIZE) { - SetPageUptodate(page); - } else if (!PagePrivate(page) && - (bp->b_flags & _XBF_PAGE_CACHE)) { - set_page_region(page, bvec->bv_offset, bvec->bv_len); - } - - if (--bvec >= bio->bi_io_vec) - prefetchw(&bvec->bv_page->flags); - - if (bp->b_flags & _XBF_PAGE_LOCKED) - unlock_page(page); - } while (bvec >= bio->bi_io_vec); - _xfs_buf_ioend(bp, 1); bio_put(bio); } @@ -1310,7 +1207,6 @@ _xfs_buf_ioapply( int offset = bp->b_offset; int size = bp->b_count_desired; sector_t sector = bp->b_bn; - unsigned int blocksize = bp->b_target->bt_bsize; total_nr_pages = bp->b_page_count; map_i = 0; @@ -1331,29 +1227,6 @@ _xfs_buf_ioapply( (bp->b_flags & XBF_READ_AHEAD) ? READA : READ; } - /* Special code path for reading a sub page size buffer in -- - * we populate up the whole page, and hence the other metadata - * in the same page. This optimization is only valid when the - * filesystem block size is not smaller than the page size. - */ - if ((bp->b_buffer_length < PAGE_CACHE_SIZE) && - ((bp->b_flags & (XBF_READ|_XBF_PAGE_LOCKED)) == - (XBF_READ|_XBF_PAGE_LOCKED)) && - (blocksize >= PAGE_CACHE_SIZE)) { - bio = bio_alloc(GFP_NOIO, 1); - - bio->bi_bdev = bp->b_target->bt_bdev; - bio->bi_sector = sector - (offset >> BBSHIFT); - bio->bi_end_io = xfs_buf_bio_end_io; - bio->bi_private = bp; - - bio_add_page(bio, bp->b_pages[0], PAGE_CACHE_SIZE, 0); - size = 0; - - atomic_inc(&bp->b_io_remaining); - - goto submit_io; - } next_chunk: atomic_inc(&bp->b_io_remaining); @@ -1367,8 +1240,9 @@ next_chunk: bio->bi_end_io = xfs_buf_bio_end_io; bio->bi_private = bp; + for (; size && nr_pages; nr_pages--, map_i++) { - int rbytes, nbytes = PAGE_CACHE_SIZE - offset; + int rbytes, nbytes = PAGE_SIZE - offset; if (nbytes > size) nbytes = size; @@ -1383,7 +1257,6 @@ next_chunk: total_nr_pages--; } -submit_io: if (likely(bio->bi_size)) { if (xfs_buf_is_vmapped(bp)) { flush_kernel_vmap_range(bp->b_addr, @@ -1393,18 +1266,7 @@ submit_io: if (size) goto next_chunk; } else { - /* - * if we get here, no pages were added to the bio. However, - * we can't just error out here - if the pages are locked then - * we have to unlock them otherwise we can hang on a later - * access to the page. - */ xfs_buf_ioerror(bp, EIO); - if (bp->b_flags & _XBF_PAGE_LOCKED) { - int i; - for (i = 0; i < bp->b_page_count; i++) - unlock_page(bp->b_pages[i]); - } bio_put(bio); } } @@ -1450,7 +1312,7 @@ xfs_buf_iowait( trace_xfs_buf_iowait(bp, _RET_IP_); if (atomic_read(&bp->b_io_remaining)) - blk_run_address_space(bp->b_target->bt_mapping); + blk_run_backing_dev(bp->b_target->bt_bdi, NULL); wait_for_completion(&bp->b_iowait); trace_xfs_buf_iowait_done(bp, _RET_IP_); @@ -1468,8 +1330,8 @@ xfs_buf_offset( return XFS_BUF_PTR(bp) + offset; offset += bp->b_offset; - page = bp->b_pages[offset >> PAGE_CACHE_SHIFT]; - return (xfs_caddr_t)page_address(page) + (offset & (PAGE_CACHE_SIZE-1)); + page = bp->b_pages[offset >> PAGE_SHIFT]; + return (xfs_caddr_t)page_address(page) + (offset & (PAGE_SIZE-1)); } /* @@ -1491,9 +1353,9 @@ xfs_buf_iomove( page = bp->b_pages[xfs_buf_btoct(boff + bp->b_offset)]; cpoff = xfs_buf_poff(boff + bp->b_offset); csize = min_t(size_t, - PAGE_CACHE_SIZE-cpoff, bp->b_count_desired-boff); + PAGE_SIZE-cpoff, bp->b_count_desired-boff); - ASSERT(((csize + cpoff) <= PAGE_CACHE_SIZE)); + ASSERT(((csize + cpoff) <= PAGE_SIZE)); switch (mode) { case XBRW_ZERO: @@ -1606,7 +1468,6 @@ xfs_free_buftarg( xfs_flush_buftarg(btp, 1); if (mp->m_flags & XFS_MOUNT_BARRIER) xfs_blkdev_issue_flush(btp); - iput(btp->bt_mapping->host); kthread_stop(btp->bt_task); kmem_free(btp); @@ -1630,15 +1491,6 @@ xfs_setsize_buftarg_flags( return EINVAL; } - if (verbose && - (PAGE_CACHE_SIZE / BITS_PER_LONG) > sectorsize) { - printk(KERN_WARNING - "XFS: %u byte sectors in use on device %s. " - "This is suboptimal; %u or greater is ideal.\n", - sectorsize, XFS_BUFTARG_NAME(btp), - (unsigned int)PAGE_CACHE_SIZE / BITS_PER_LONG); - } - return 0; } @@ -1653,7 +1505,7 @@ xfs_setsize_buftarg_early( struct block_device *bdev) { return xfs_setsize_buftarg_flags(btp, - PAGE_CACHE_SIZE, bdev_logical_block_size(bdev), 0); + PAGE_SIZE, bdev_logical_block_size(bdev), 0); } int @@ -1665,41 +1517,6 @@ xfs_setsize_buftarg( return xfs_setsize_buftarg_flags(btp, blocksize, sectorsize, 1); } -STATIC int -xfs_mapping_buftarg( - xfs_buftarg_t *btp, - struct block_device *bdev) -{ - struct backing_dev_info *bdi; - struct inode *inode; - struct address_space *mapping; - static const struct address_space_operations mapping_aops = { - .sync_page = block_sync_page, - .migratepage = fail_migrate_page, - }; - - inode = new_inode(bdev->bd_inode->i_sb); - if (!inode) { - printk(KERN_WARNING - "XFS: Cannot allocate mapping inode for device %s\n", - XFS_BUFTARG_NAME(btp)); - return ENOMEM; - } - inode->i_ino = get_next_ino(); - inode->i_mode = S_IFBLK; - inode->i_bdev = bdev; - inode->i_rdev = bdev->bd_dev; - bdi = blk_get_backing_dev_info(bdev); - if (!bdi) - bdi = &default_backing_dev_info; - mapping = &inode->i_data; - mapping->a_ops = &mapping_aops; - mapping->backing_dev_info = bdi; - mapping_set_gfp_mask(mapping, GFP_NOFS); - btp->bt_mapping = mapping; - return 0; -} - STATIC int xfs_alloc_delwrite_queue( xfs_buftarg_t *btp, @@ -1728,12 +1545,14 @@ xfs_alloc_buftarg( btp->bt_mount = mp; btp->bt_dev = bdev->bd_dev; btp->bt_bdev = bdev; + btp->bt_bdi = blk_get_backing_dev_info(bdev); + if (!btp->bt_bdi) + goto error; + INIT_LIST_HEAD(&btp->bt_lru); spin_lock_init(&btp->bt_lru_lock); if (xfs_setsize_buftarg_early(btp, bdev)) goto error; - if (xfs_mapping_buftarg(btp, bdev)) - goto error; if (xfs_alloc_delwrite_queue(btp, fsname)) goto error; btp->bt_shrinker.shrink = xfs_buftarg_shrink; @@ -1955,7 +1774,7 @@ xfsbufd( count++; } if (count) - blk_run_address_space(target->bt_mapping); + blk_run_backing_dev(target->bt_bdi, NULL); } while (!kthread_should_stop()); @@ -2003,7 +1822,7 @@ xfs_flush_buftarg( if (wait) { /* Expedite and wait for IO to complete. */ - blk_run_address_space(target->bt_mapping); + blk_run_backing_dev(target->bt_bdi, NULL); while (!list_empty(&wait_list)) { bp = list_first_entry(&wait_list, struct xfs_buf, b_list); diff --git a/fs/xfs/linux-2.6/xfs_buf.h b/fs/xfs/linux-2.6/xfs_buf.h index cbe65950e524..a9a1c4512645 100644 --- a/fs/xfs/linux-2.6/xfs_buf.h +++ b/fs/xfs/linux-2.6/xfs_buf.h @@ -61,30 +61,11 @@ typedef enum { #define XBF_DONT_BLOCK (1 << 16)/* do not block in current thread */ /* flags used only internally */ -#define _XBF_PAGE_CACHE (1 << 17)/* backed by pagecache */ #define _XBF_PAGES (1 << 18)/* backed by refcounted pages */ #define _XBF_RUN_QUEUES (1 << 19)/* run block device task queue */ +#define _XBF_KMEM (1 << 20)/* backed by heap memory */ #define _XBF_DELWRI_Q (1 << 21)/* buffer on delwri queue */ -/* - * Special flag for supporting metadata blocks smaller than a FSB. - * - * In this case we can have multiple xfs_buf_t on a single page and - * need to lock out concurrent xfs_buf_t readers as they only - * serialise access to the buffer. - * - * If the FSB size >= PAGE_CACHE_SIZE case, we have no serialisation - * between reads of the page. Hence we can have one thread read the - * page and modify it, but then race with another thread that thinks - * the page is not up-to-date and hence reads it again. - * - * The result is that the first modifcation to the page is lost. - * This sort of AGF/AGI reading race can happen when unlinking inodes - * that require truncation and results in the AGI unlinked list - * modifications being lost. - */ -#define _XBF_PAGE_LOCKED (1 << 22) - typedef unsigned int xfs_buf_flags_t; #define XFS_BUF_FLAGS \ @@ -100,12 +81,10 @@ typedef unsigned int xfs_buf_flags_t; { XBF_LOCK, "LOCK" }, /* should never be set */\ { XBF_TRYLOCK, "TRYLOCK" }, /* ditto */\ { XBF_DONT_BLOCK, "DONT_BLOCK" }, /* ditto */\ - { _XBF_PAGE_CACHE, "PAGE_CACHE" }, \ { _XBF_PAGES, "PAGES" }, \ { _XBF_RUN_QUEUES, "RUN_QUEUES" }, \ - { _XBF_DELWRI_Q, "DELWRI_Q" }, \ - { _XBF_PAGE_LOCKED, "PAGE_LOCKED" } - + { _XBF_KMEM, "KMEM" }, \ + { _XBF_DELWRI_Q, "DELWRI_Q" } typedef enum { XBT_FORCE_SLEEP = 0, @@ -120,7 +99,7 @@ typedef struct xfs_bufhash { typedef struct xfs_buftarg { dev_t bt_dev; struct block_device *bt_bdev; - struct address_space *bt_mapping; + struct backing_dev_info *bt_bdi; struct xfs_mount *bt_mount; unsigned int bt_bsize; unsigned int bt_sshift; @@ -139,17 +118,6 @@ typedef struct xfs_buftarg { unsigned int bt_lru_nr; } xfs_buftarg_t; -/* - * xfs_buf_t: Buffer structure for pagecache-based buffers - * - * This buffer structure is used by the pagecache buffer management routines - * to refer to an assembly of pages forming a logical buffer. - * - * The buffer structure is used on a temporary basis only, and discarded when - * released. The real data storage is recorded in the pagecache. Buffers are - * hashed to the block device on which the file system resides. - */ - struct xfs_buf; typedef void (*xfs_buf_iodone_t)(struct xfs_buf *);